Emergency clinicians who order the fewest CT scans for lung clots have nearly twice the rate of clots that leave the hospital undiagnosed, according to a study of 4,180,159 emergency department visits published September 22, 2026 in JAMA Network Open. The doctors who tested least found a clot on 9.6% of their scans, the best hit rate in the study, and also had the highest rate of potentially missed clots: 6.6 per 100,000 visits against 3.5 per 100,000 for the doctors who tested most.
Here is what that means if you are the patient. The doctor who looks most disciplined about ordering scans is not automatically the one most likely to catch a clot in your lung. That is not a reason to demand a CT scan. It is a reason to ask, before you are discharged, whether a lung clot was considered and how it was ruled out.
Whether a test catches risk that routine testing misses comes up outside the ER too, and one study put a glucose monitor up against ordinary blood tests.
Key takeaways
- Clinicians who tested least for lung clots found one on 9.6% of their CT scans, against 6.4% for those who tested most.
- Potentially missed lung clots ran at 6.6 per 100,000 emergency visits among the least-testing clinicians and 3.5 per 100,000 among the most-testing ones.
- Across 29 Michigan emergency departments, 189 of 13,530 acute lung clots, or 1.4%, were flagged as potentially missed.
What the study found
The retrospective cohort study sampled about 5.1 million visits to 29 Michigan emergency departments from January 1, 2023 through November 30, 2025. After exclusions, including patients who left against medical advice or before treatment was finished, and scans done in children, trauma activations, pregnant patients, or for reasons other than a suspected lung clot, 4,180,159 visits and 747 clinicians were left to analyze. Median patient age was 52.
The researchers split a lung-clot workup into three decisions: whether to test at all, whether to go straight to a CT scan or start with a D-dimer blood test, and whether to scan anyway after a negative D-dimer. Only the first moved the numbers much.
Ranked by how often they tested, the bottom fifth of clinicians found an acute clot on 9.6% of their CT scans (95% CI, 9.0% to 10.1%). The top fifth found one on 6.4% (95% CI, 5.5% to 7.3%). That is a 3.2 percentage point gap, and a true gap of 2.1 to 4.2 points is very likely.
Missed clots were counted indirectly, from patients who returned to an ER within 10 days and were then diagnosed: 189 potentially missed clots out of 13,530 acute cases, or 1.4%, clustered among the clinicians who tested least.
Dr. Kumar’s take
Diagnostic yield is a hospital quality metric, not a measure of whether your doctor got it right. It is the percentage of scans that find something, and there is a trivial way to raise it: scan fewer people. A clinician with excellent judgment and one who never thinks of the diagnosis produce the same number on that report card.
For a decade the loudest message in emergency medicine has been that the ER scans too much for lung clots. That was not wrong. This study puts a price on the correction: about 3 more potentially missed clots per 100,000 visits at the low-testing end.
Those numbers are small, and so is the 1.4% missed rate. A 10-day return visit infers a miss rather than proving one, and some of those 189 patients may have formed a clot after going home.
The tradeoff runs the other way too. The authors calculated that if everyone tested as sparingly as the lowest group, an estimated 41,966 CT scans, about 26.4%, would have been avoided, and scanning carries radiation, contrast dye, and incidental findings. Yield alone cannot locate the right line, because the most efficient looking practice and the most dangerous one produce the same chart.
This is one state and one three-year window, and clinicians were sorted by their own behavior rather than assigned a strategy.
What it means for you
Lung clots are treatable and dangerous when missed. The situations that raise the odds are worth knowing: recent surgery, a long stretch of immobility, a long flight, swelling or pain in one leg, a cancer diagnosis, estrogen-containing medication, or a past clot.
If any of that applies and you turn up with unexplained shortness of breath or chest pain, ask plainly: was a blood clot in my lung considered, and what ruled it out? A D-dimer blood test is often the first step, and the study treats a negative age-adjusted D-dimer as a reason not to go on to a scan.
What this does not license is arriving with a scan already decided on. Who needs imaging depends on your symptoms, history, and exam, and that call belongs with your doctor. The same question comes up in cardiology, where one analysis sorted out who actually benefits from a coronary artery calcium scan.